[ad_1]
If it was not enough artificial intelligence that beats humans to Dota 2 to worry, here are even developers of DeepMind ventured into something like that and after beating Go men they decided that it was time for artificial intelligence to engage in video games, especially with a modified version of Quake III Arena in Capture The Flag mode.
The goal is always the same: ] find new ways of cooperation between AI and develop solutions for learning . Here the developers of DeepMind have chosen a different path than the one followed by the developers of OpenAI
Rather than setting up a kind of "super brain" able to evaluate time in time it is necessary to encourage the pursuit of the objectives of the AI or the team to always get the best possible result, special algorithms have been developed to allow the four AI involved to "see" and to "feel".
In fact, real-time badysis of the game scenario just like us humans, and decides how to proceed by also coordinating with The choices made by other AIs The whole is based on a system of "signal reinforcement" a technique that through the badysis of the errors made can constantly improve. The AI that "improves the evil" could in the future be used not for military purposes, for the construction of combat robots for example, but to develop useful solutions also in the civilian field, such as that drones or robots capable of intervening effectively in emergency scenarios such as cataclysms or fires, for example.
Source link
